Three different methods have been used to determine the sputtering yield (S) for 2–30 keV Kr ions on the amorphous anodic oxides Nb2O5, Ta2O5, and WO3. The methods were based on (a) the change of interference colors due to ion bombardment, (b) the perforation of a film of free oxide by an intense ion beam, and (c) the loss of weight of the target. Although S for Kr–WO3 was larger than expected, S for Kr–Nb2O5 and Kr–Ta2O5 was similar to what has been found for other oxides. For example, at 4 keV, S lies between 1.1 and 2.4 atoms/ion for Ar or Kr bombardment of Al2O3, Nb2O5, SiO2, Ta2O5, TiO2, and UO2, whereas it is 6.2 atoms/ion for WO3. The results were interpreted first with a modified form of the theory of Rol et al., in which S is equated to K′E/Rm, where K′ is a constant in Å/keV and Rm is the experimental median range. A good fit between experiment and theory was obtained for K′ equal to 25 for Nb2O5, 13 for Ta2O5, and 51 for WO3. The recently formulated theory of Sigmund, which is in principle completely analytical, was also considered. The observed values of S could be made compatible with the predictions of the theory by taking the following values for Eb, the surface binding energy (units of kcal/gram atom): 100–150 for Nb2O5, 200–250 for Ta2O5, and 45–60 for WO3, We would point out that these latter values are similar to the heats of atomization.